Continental warming preceding the Paleoceneâ•fiEocene thermal maximum

Marine and continental records1 show an abrupt negative shift in carbon isotope values at ~55.8 Myr ago. This carbon isotope excursion (CIE) is consistent with the release of a massive amount of isotopically light carbon into the atmosphere and was associated with a dramatic rise in global temperatures termed the Paleocene–Eocene thermal maximum (PETM). Greenhouse gases released during the CIE, probably including methane, have often been considered the main cause of PETM warming. However, some evidence from the marine record suggests that warming directly preceded the CIE,2, 3, 4 raising the possibility that the CIE and PETM may have been linked to earlier warming with different origins. Yet pre-CIE warming is still uncertain. Disentangling the sequence of events before and during the CIE and PETM is important for understanding the causes of, and Earth system responses to, abrupt climate change. Here we show that continental warming of about 5 °C preceded the CIE in the Bighorn Basin, Wyoming. Our evidence, based on oxygen isotopes in mammal teeth (which reflect temperature-sensitive fractionation processes) and other proxies, reveals a marked temperature increase directly below the CIE, and again in the CIE. Pre-CIE warming is also supported by a negative amplification of δ13C values in soil carbonates below the CIE. Our results suggest that at least two sources of warming—the earlier of which is unlikely to have been methane—contributed to the PETM. Abrupt warming during the PETM coincided with a profound reorganization of mammalian fauna as the first representatives of several modern groups dispersed among the northern continents.5 In North America, some plant species shifted northward by as much as 1,500 km (reference 6). In the marine realm, the PETM is marked by the extinction of benthic foraminifers, and increased CO2 levels caused a shoaling of the lysocline and dissolution of carbonates.7 Estimates of marine warming in surface ocean water during the PETM range from 4 °C to 9 °C. Previous estimates of continental warming during the PETM range from ~5 °C using leaf-margin analyses of palaeofloras,8 to ~2–7 °C using oxygen isotopes in mammalian tooth enamel9 (δOE) or soil carbonates.10 These estimates averaged a large interval of late Paleocene time, however, and with the exception of the soil carbonates, lacked the resolution to detect a change in climate directly below the CIE. Here we present the highest-resolution δOE record yet compiled and use this record along with published data to infer the timing of PETM climate change. We use oxygen isotopes in mammalian tooth enamel (δOE) from both phosphate (δOEP) and carbonate (δOEC) as proxies for surface water composition (δOSF). Surface water should broadly reflect changes in meteoric water.11, 12 Oxygen values in carbonates and phosphates are determined by the isotopic composition of the fluids from which these minerals precipitate and by the temperature at precipitation. Because mammals maintain a constant body temperature of ~37 °C, δ18O values of mammalian tooth enamel (δOE) reflect the composition of body water. The composition of body water, in turn, is determined by that of ingested water, atmospheric oxygen (which remains constant), metabolic water from food, and by physiological processes such as panting and sweating.12, 13 To minimize possible variability caused by differences in drinking habits, diet, and physiology among taxa, we use teeth only from closely related species of the archaic ungulate Phenacodus. Because most medium to large mammals in temperate climates follow seasonal birthing cycles, a given tooth position should form during the same season in all individuals of a species. Thus, to minimize error from seasonal variability in the δ18O of precipitation we use only a single tooth position (upper or lower third molars). A large decrease in the δCEC values in Phenacodus of 3.7‰ occurs from the ~12,000-year (12-kyr) interval (6 m thick) directly below the CIE to within the CIE (Figure 1a). A previous study found a similar decrease of ~4‰ in δCEC during the CIE.14 Carbon isotope values in mammals are a direct reflection of their diet with a predictable amount of metabolic 13C-enrichment.15 Thus, the teeth of herbivores reflect the δ13C values of local vegetation, which in turn reflects the value of atmospheric CO2 and local environmental conditions such as canopy development and water availability.16 Phenacodus was clearly an omnivore on the basis of its dentition, but even the δCEC values of carnivores closely follow those of local vegetation through their prey species. Large increases in δOE occur directly below the CIE and again in the CIE (Figure 1b and c and Figure 2b and c). Significant increases (P ≤ 0.05) in mean δOEP and δOEC of 1.3‰ and 1.7‰, respectively, occur from the upper Paleocene to the 12-kyr interval directly below the CIE for underlying bin sizes of five or more samples (see Supplementary Table S1 for statistics and bin sizes). These increases are followed by increases of 1.9‰ and 1.3‰ in δOEP and δOEC, respectively, from the 12-kyr interval to within the CIE. However, of these last two increases, only δOEP is significant with 95% confidence and only with an underlying bin size of seven or more samples. The low significance in δOEC is due to small sample size and high variability in the CIE. The observed increases are all consistent with warming starting before and continuing into the CIE, based on modern meteoric water/temperature relationships.17 A strong positive correlation exists today between mean annual temperature (MAT) and δ18O values in precipitation at mid and high latitudes,17 which is reflected in local δOSF. However, this relationship is dependent on the modern latitudinal temperature gradient, and it may not be reliable for estimating MAT change in the early Cenozoic when the gradient may have been lower than today.18 To address this problem we calculated δOSF/ MAT slopes specific to the early Eocene of the Bighorn Basin using published MAT estimates from fossil leaves as a proxy for temperature, and values for δOEC (reference 16) and the δ18O of hematite8 as proxies for δOSF. This yielded δOSF/MAT slopes of 0.39 ± 0.24 and 0.36 ± 0.11 (‰ per °C) (all error in this paper is reported as 1.96 × s.e.m., that is, 95% confidence) from enamel and hematite, respectively (Figure 3 and Supplementary Information). These estimates are in good agreement with each other and suggest a lower meteoric water/MAT slope than the modern one for Published in Nature 467 (October 21, 2010), pp. 955–958; doi:10.1038/nature09441 Copyright © 2010 Macmillan Publishers Ltd. Used by permission. Submitted May 3, 2010; accepted August 17, 2010; published online October 20, 2010. Continental warming preceding the Paleocene–Eocene thermal maximum Ross Secord,1, 2, 3 Philip D. Gingerich,2 Kyger C. Lohmann,2 and Kenneth G. MacLeod 4 1. Department of Earth and Atmospheric Sciences, University of Nebraska, Lincoln, Nebraska 68588, USA 2. Department of Geological Science, University of Michigan, Ann Arbor, Michigan 48109, USA 3. Florida Museum of Natural History, Gainesville, Florida 32611, USA 4. Department of Geological Sciences, University of Missouri, Columbia, Missouri 65211, USA Corresponding author — Ross Secord, [email protected]